Abstract
Context
Cushing’s syndrome (CS) impairs quality of life (QoL) and mood. Prospective real-life data on posttreatment recovery and predictors of improvement are limited.
Objectives
Evaluate changes in QoL, depression, and anxiety in patients with CS, before and after biochemical control, and identify predictors of clinically meaningful improvement.
Design and Setting
Prospective observational study at a tertiary center.
Patients
Sixty-seven patients with endogenous CS (60 pituitary, 7 adrenal) were assessed with active disease and again after achieving biochemical control through surgery and/or medication.
Outcomes
Patient-reported outcomes included CushingQoL, Beck Depression Inventory-II (BDI-II), and State-Trait Anxiety Inventory (STAI).
Results
Mean and longest follow-up was 2.3 and 11.5 years, respectively. Treatment led to improvements in mean scores across all domains (QoL: +18.2 ± 20.9, BDI-II: −6.8 ± 8.6, STAI-State: −9.6 ± 12.5, STAI-Trait: −8.6 ± 12.6; all P < .001). However, a minimal important difference was achieved in 64.6% for QoL, 67.9% for BDI-II, and 53.2% and 52.8% for STAI subscales. After multivariable analysis, QoL improvements were predicted by lower baseline body mass index, pretreatment symptoms ❤ years, postoperative hydrocortisone replacement >6 months, and normal follow-up late-night salivary cortisol (LNSC). Depression improvements were predicted by symptoms ❤ years, normal follow-up LNSC, and surgical treatment. Anxiety improvements were predicted by younger age and >6 months postoperative hydrocortisone. Depression improved more gradually than QoL and anxiety.
Conclusion
Although effective treatment improves mood and QoL in CS, clinically meaningful recovery is variable and incomplete for some patients. Our findings highlight the need to limit diagnostic delay and provide comprehensive posttreatment care that includes normalization of cortisol circadian rhythm.
Endogenous Cushing’s syndrome (CS) is a rare disorder characterized by chronic cortisol excess, most commonly due to an ACTH-secreting pituitary tumor [Cushing disease (CD)], followed by a cortisol-secreting adrenal adenoma and ectopic ACTH production due to a nonpituitary tumor (1). CS is associated with multiple comorbidities including diabetes, obesity, hypertension, immune suppression, osteoporosis, and cardiovascular disease, among others (2). Apart from these, patients face a spectrum of neuropsychiatric disturbances including depression, anxiety, mania, sleep disorders, and even psychosis. These comorbidities significantly disturb quality of life (QoL) and may persist long after treatment (3-7).
As with many rare diseases, CS remains incompletely understood, and patients experience impaired disease perception, information gaps, and isolation. In this context, patient-reported outcomes (PROs) have become useful instruments to clarify these gaps and guide patient-centered care. Disease-specific tools (CushingQoL, Tuebingen CD-25) and generic mood scales (Beck Depression Inventory, State-Trait Anxiety Inventory [STAI; including State (STAI-S) and Trait (STAI-T), Hospital Anxiety and Depression Scale] have established impairments in QoL and mood both during active disease and in remission (4, 8-11).
Although improvements are noted with treatment, recovery does not seem to be complete. Studies have reported persistently reduced QoL compared to the general population and the presence of depressive symptoms even 12 months postoperatively (4, 9). Findings regarding anxiety are less consistent: while some studies did not support the increased prevalence of anxiety in patients with active CS compared to the general population (12), others reported higher anxiety traits among patients with CS (during active disease and in remission) (13, 14) with steady improvement at 6- and 12-month follow-up (15). Clinical trials with adrenal steroidogenesis inhibitors or pasireotide demonstrated that effective biochemical control can improve QoL and depression (16-18). However, it is unclear whether these improvements are clinically significant and if patients achieve normal QoL and depression scores.
The role of PROs in assessing recovery during the treatment journey of patients with CS has not been clearly established, and QoL and mood trajectories remain unclear, largely due to small samples, limited follow-up, and cross-sectional designs. Among available prospective studies using PROs in CS, only 3 (2 evaluating pasireotide and 1 osilodrostat) reported the proportion of patients who met the minimal important difference (MID), which is the score change reflecting a clinically meaningful improvement (17-19), while others have only reported statistically significant changes in mean score, an important but possibly less clinically relevant outcome (20-22). Real-world clinical management adds further complexity: postoperative glucocorticoid replacement, potential glucocorticoid-withdrawal symptoms, and 20% to 30% recurrence rates after initial surgical “cure” all suggest that, for many patients, recovery may follow a nonlinear course. To date, no clinical practice prospective study has systematically assessed QoL and mood across multiple timepoints, compared surgical and medical strategies within a single cohort, and limited inclusion to patients who achieved biochemical remission or control for at least 6 months. Therefore, the aims of this study were to evaluate changes in QoL, depression, and anxiety in a clinical practice cohort of patients with CS before and over time after biochemical control, report achievement rates of MID, and identify predictors of clinically meaningful improvement.
Methods
Study Design
This study includes prospective data from patients enrolled in an ongoing observational cohort study, which since 2017 enrolls patients with endogenous CS at Memorial Sloan Kettering Cancer Center (MSKCC) [prior to 2017, enrollment took place at Mount Sinai (2012-2017)]. In this protocol, CS patients being treated at the MSKCC Pituitary and Skull Base Tumor Center are enrolled at any point in their treatment journey and prospectively followed over time after surgical, medical, and/or radiation treatment. At each study visit, a detailed medical history and biochemical and clinical data are collected according to standard of care. Patients also complete validated psychological and QoL assessments.
The current analysis includes a cohort of 67 patients with CS: 60 with pituitary and 7 with adrenal CS. Each patient completed a baseline (active disease) visit and at least 1 follow-up visit after achieving surgical remission or endocrine control due to medical therapy.
From the total of 67 patients, we analyzed 73 distinct baseline-to-follow-up case pairs. Six patients experienced recurrence after surgery or were inadequately controlled while on medical therapy after their initial follow-up visit and underwent a subsequent change in treatment strategy. These instances were treated as separate case pairs when needed, enabling comparison of different treatment approaches. When analyzing for a single follow-up, visits were grouped by time: group 1 (G1): 6 months, group 2 (G2): 12-18 months, and group 3 (G3): 24 or more months posttreatment. Each patient contributed to 1 or multiple groups based on the number of their study visits. For patients with multiple visits receiving different treatments throughout the current study, each follow-up visit was categorized based on time since the most recent intervention to ensure that we assessed outcomes according to the duration of biochemical control. For patients who underwent surgery, the follow-up interval was calculated from the date of surgery; for those on medical therapy, it was calculated from the start of medication. In the subanalysis comparing treatment- or demographic-related score changes, the most recent available follow-up was used in each case. At each visit patients completed at least 1 of the following: Cushing QoL, Beck Depression Inventory-II (BDI-II), or STAI-S and STAI-T.
For multiple follow-up visits during remission or treatment, 28 patients were evaluated. For this subgroup, we examined their whole trajectory over time. We then stratified this subgroup by total follow-up duration (<2 years vs ≥2 years) and assessed for significant differences between these 2 categories where applicable.
For the baseline visit, ACTH-dependent pituitary and ACTH-independent adrenal Cushing’s was confirmed according to Endocrine Society guidelines (23). Surgical remission was defined as postoperative serum cortisol <5 μg/dL (<138 nmol/L) and requirement of glucocorticoid replacement, according to the Endocrine Society’s guidelines and the Pituitary Society’s recent consensus statement (24, 25). For patients managed medically, endocrine control was defined as normalization of 24-hour urinary free cortisol (UFC) and based on clinical review and assessment by E.B.G.
The study was approved by the institutional review board at MSKCC. All subjects gave written informed consent before participation.
Outcome Measurements
Cushing QoL
The Cushing QoL is a validated disease-specific questionnaire consisting of 12 questions on a 5-point scale ranging from “always” to “never” (for 10 questions) or “very much” to “not at all” (for 2 questions). Total score ranges from 12 to 60. This is converted to a 0 to 100 scale, with 0 indicating the worst and 100 the best QoL. It evaluates physical and psychological issues and can also be scored through these 2 distinct subscales. MID is defined as an increase of ≥10.1 (26).
BDI-II
The BDI-II is a validated 21-item patient-reported questionnaire. Patients self-rate each item on a scale from 0 to 3 based on how they were feeling during the past 2 weeks. Total score ranges from 0 (best) to 63 (worst); scores from 0 to 13 indicate no or minimal depression; 14 to 19, mild depression; 20 to 28, moderate depression; and 29 to 63, severe depression. MID is defined as a 20% reduction from baseline score (27, 28).
STAI
The STAI is an instrument with 2 subscales: State anxiety (STAI-S), which reflects the present moment, and Trait anxiety (STAI-T), which assesses a stable tendency toward anxiety. Both subscales consist of 20 items scored from 0 to 3. Total scores range from 0 to 60, with higher scores indicating greater anxiety. Prior studies suggest a change of 0.5× SDs—or approximately 5 to 10 points—as a reasonable threshold for MID. In our study, we defined the MID at 7 points, based on observed SD of change at 12.5 for STAI-S and 12.6 for STAI-T (29).
In this study, all score changes from baseline to follow-up were reported as positive values to uniformly represent improvement across measures. For BDI-II and STAI where higher scores indicate worse outcomes, the direction of change was inverted for consistency.
Hormone Assays
Hormone testing was performed at either the MSKCC clinical laboratory or external laboratories (Quest Diagnostics, Labcorp, Mayo Clinic Laboratories). Plasma ACTH was measured using Tosoh immunoassay [RRID:AB_2783633; normal range (NR): 7.4-64.3 pg/mL (1.6-14.2 pmol/L); MSKCC or 6 to 50 pg/mL (1.3-11.0 pmol/L); QuestDiagnostics] or electrochemiluminescence immunoassay [RRID:AB_3678556; NR: 7.2-63.3 pg/mL (1.6-13.9 pmol/L); LabCorp, Mayo Clinic Laboratories]. Serum cortisol was measured via either immunoassay [RRID:AB_2802133; NR: 4-22 µg/dL (110-607 nmol/L); QuestDiagnostics or 7-25 µg/dL (193-690 nmol/L); Mayo Clinic Laboratories], electrochemiluminescence immunoassay [RRID:AB_2802131; NR: 6.2-19.4 µg/dL; (171-535 nmol/L); LabCorp], or liquid chromatography–tandem mass spectrometry [LC-MS/MS; NR: 5-25 µg/dL (138-690 nmol/L)]. UFC was measured using LC-MS/MS [NR: 3.5-45 µg/24 hours (9.7-124 nmol/24 hours); MSKCC, Mayo Clinic Laboratories or 3.0 to 50 µg/24 hours (8.3-138 nmol/24 hours); Quest Diagnostics, LabCorp]. Late-night salivary cortisol (LNSC) was assessed via LC-MS/MS [NR: ≤ 0.09 µg/dL (2.5 nmol/L); QuestDiagnostics, LabCorp or <100 ng/dL (27.6 nmol/L); MSKCC, Mayo Clinic Laboratories]. LNSC values were analyzed categorically (normal vs abnormal), and patients were asked to provide 2 LNSC samples on separate evenings. Abnormal LNSC was defined as at least 1 value above the upper limit of normal for the assigned laboratory.
Comorbidities
Diabetes mellitus (DM) was defined by any of the following: hemoglobin A1c (HbA1c) > 6.4%, fasting blood glucose (FBG) ≥ 126 mg/dL (7.0 mmol/L), or use of at least 1 antidiabetic medication. Pre-DM was defined as HbA1c between 5.7% and 6.4% or FBG between 100 and 125 mg/dL (5.6-6.9 mmol/L). Women taking metformin for polycystic ovary syndrome were classified as nondiabetic only if their HbA1c and FBG values both before metformin initiation and at the time of CS diagnosis remained within the normal range. Hypertension was defined as systolic blood pressure ≥ 130 mmHg, diastolic blood pressure ≥ 80 mmHg, or use of any antihypertensive medication.
Statistical Analysis
Analyses were conducted using IBM SPSS for Windows (version 29.0, IBM Corp.). Data normality was assessed by the Shapiro–Wilk test. Descriptive statistics were used for demographic and clinical characteristics. Normally distributed data were compared by Student’s t-test and nonnormally distributed variables with the Mann–Whitney U-test. Paired T-tests were conducted to study mean changes from baseline to a single follow-up visit. For categorical characteristics and the MID, we calculated the achievement rates and used Pearson’s chi-square for comparisons where applicable. For patients with more than 2 follow-up visits ANOVA (repeated measures) was applied for the trajectory of each measurement over time. To identify predictors of improvement, univariable linear regression models for score change and logistic regression for MID achievement were performed using baseline visit and longest follow-up visit for each patient. Variables with P ≤ .10 or of clinical relevance were then entered into multivariable regression models—again, linear regression for score change and logistic regression for MID achievement—where each predictor was separately evaluated, adjusting for age, sex, and baseline score. Correlation analyses were performed using Pearson or Spearman correlation coefficients for data with normal or abnormal distribution, respectively. Correlation coefficients (r) were interpreted as follows: values between 0.0 and ±0.3: weak, between ±0.3 and ±0.7: moderate, and between ±0.7 and ±1.0: strong relationships. All statistical tests were 2-sided, and results were considered significant with P ≤ .05.
Results
Study Participants
From a cohort of 226 endogenous CS and silent ACTH tumor patients enrolled in our ongoing MSKCC prospective cohort study, we identified patients who had a baseline visit with active hypercortisolism, who had at least 1 follow-up visit while in surgical remission or medical control, and who had completed at least 1 of the evaluated questionnaires correctly. After excluding patients with silent ACTH tumors, those with missing data, and follow-up visits that did not meet remission criteria, we included 67 patients (56 females, 11 males) with a mean baseline age of 42.3 ± 13.1 years. Among these patients, 60 had CD and 7 had adrenal CS.
Further patient demographic information is shown in Tables 1 and 2.
Table 1.
Demographics and baseline characteristics
| Demographic variable |
n = 67 patients |
| Age, years |
|
| Mean (SD) |
42.3 (13.1) |
| Range |
20-75 |
| Sex, n (%) |
|
| Female |
56 (83.6) |
| CS subtype, n (%) |
|
| CD |
60 (89.6) |
| Adrenal CS |
7(10.4) |
| Race, n (%) |
|
| White |
50 (74.6) |
| Black/African American |
8 (11.9) |
| Asian |
2 (3.0) |
| Other/unknown |
7 (10.4) |
| 24-hour UFC |
|
| Mean (SD) |
391.5 (1471) µg/24 hours,
1080 (4060) nmol/24 hours |
| Median (IQR) |
135.0 (82.7-220.0) µg/24 hours, 372 (228-607) nmol/24 hours |
| Range (min-max) |
29-12 346 µg/24 hours, 80-34 053 nmol/24 hours |
| LNSC, n (%) |
|
| Normal |
3 (4.5) |
| Abnormal |
59 (88.1) |
| NA |
5 (7.5) |
| Plasma ACTH |
|
| Mean (SD) |
70.7 (64.1) pg/mL, 15.6 (14.1) pmol/L |
| Median (IQR) |
56.0 (42.0-83.8) pg/mL, 12.3 (9.2-18.4) pmol/L |
| Range (min-max) |
11-416 pg/mL (2.4-91.5 pmol/L) |
| Prior recurrence at baseline, n (%) |
16 (23.9) |
| Prior transsphenoidal surgery, n (%) |
16 (23.9) |
| 1 |
9 (13.4) |
| 2 |
7(10.4) |
Abbreviations: CD, Cushing disease; CS, Cushing’s syndrome; IQR, interquartile range; LNSC, late-night salivary cortisol; NA, not available; UFC, urinary free cortisol.
Table 2.
Baseline and follow-up data
|
Baseline |
Longest follow-up |
P-value |
| BMI (kg/m2) |
|
|
|
| Mean (SD) |
33.2 (7.6) |
30.6 (8.5) |
<.001 |
| Median (IQR) |
31.6 (26.8-37.3) |
29.3 (25.3-34.8) |
|
| LNSC, n (%) |
|
|
<.001 |
| Normal |
3 (4.5) |
30 (44.7) |
|
| Abnormal |
59 (88.1) |
16 (23.8) |
|
| NA |
5 (7.5) |
21 (31.3) |
|
| DM, n (%) |
|
|
<.001 |
| DM |
28 (41.8) |
13 (19.4) |
|
| Pre-DM |
15 (22.4) |
9 (13.4) |
|
| Hypertension, n (%) |
55 (82.1) |
35 (53.7) |
<.001 |
| HbA1C (%) |
|
|
<.001 |
| Total mean (SD) |
6.5 (1.8) |
5.7 (0.9) |
|
| DM/pre-DM mean (SD) |
6.9 (1.8) |
6.1 (1.0) |
|
| Antidiabetic medications, n (%) |
20 (29.9) |
(22.4) |
|
| 1 |
12 (17.9) |
(13.4) |
|
| 2 |
1 (1.5) |
(3.0) |
|
| 3 |
3 (4.5) |
(1.5) |
|
| Insulin |
4 (6.0) |
3 (4.5) |
|
| Antihypertensive medications, n (%) |
34 (50.7) |
(37.3) |
|
| 1 |
15 (22.4) |
(19.4) |
|
| 2 |
10 (14.9) |
(11.9) |
|
| ≥3 |
9 (13.4) |
4 (6.0) |
|
| Other medications, n (%) |
|
|
|
| Antidepressants |
10 (14.9) |
13 (19.4) |
|
| Anxiolytics |
12 (17.9) |
12(17.9) |
|
| Pain medications |
16 (23.9) |
23 (34.3) |
|
| Sleep medications |
16 (23.9) |
21 (31.3) |
|
| Treatment at most recent follow-up,a n (%) |
|
|
|
| Transsphenoidal surgery |
|
44 (65.7) |
|
| Medical therapy |
|
18 (26.9) |
|
| Bilateral adrenalectomy |
|
3 (4.5) |
|
| Radiation therapy |
|
1 (1.5) |
|
| Adrenalectomy (adrenal CS) |
|
7 (10.4) |
|
Abbreviations: BMI, body mass index; CS, Cushing’s syndrome; DM, diabetes mellitus; HbA1c, hemoglobin A1c; IQR, interquartile range; LNSC, late-night salivary cortisol.
a“n” refers to number of separate baseline-to-follow-up cases.
In total, there were 46 visits in G1, 31 in G2, and 24 in G3. At the most recent follow-up of each case, there were 24 visits in G1, 25 in G2, and 24 in G3.
The mean (range) duration from baseline to most recent follow-up was 28.3 (5-138) months in the overall cohort. The mean (range) follow-up duration since the most recent treatment was 6.3 (4-9) months for G1, 12.7 (10-18) months for G2, and 43.7 (23-120) months for G3. At their final follow-up visit, 44 patients (65.7%) achieved remission after transsphenoidal surgery (TSS), 18 (26.9%) were under medical control, 3 (4.5%) underwent bilateral adrenalectomy (BLA), 1 (1.5%) received radiation therapy (RT), and the 7 (10.4%) patients with adrenal CS underwent unilateral adrenalectomy (Table 2).
The following additional treatments were administered between this study’s baseline visit and longest follow-up: among the 44 patients treated with TSS at their latest follow-up, 1 underwent an additional TSS and 1 received medical therapy prior to TSS. Of the 18 medically managed patients at last follow-up, 8 (44.4%) had previously undergone TSS (3 of whom had 2 TSSs), and 2 of these 8 additionally received at least 1 different medication before switching to the 1 recorded at their last follow-up. Two (11.1%) other patients received 2 sequential medications before the final 1 at follow-up, and 1 (5.6%) patient was on a block-and-replace regimen with hydrocortisone (HC) after 2 TSSs and BLA. The complete treatment journey of patients on medical therapy, before and after entering the study, is shown in Fig. 1. Among the patients who underwent BLA at last follow-up, 1 had 2 prior TSSs, 1 had a sin1 gle prior TSS and received medical therapy and had 2 TSSs and received medical therapy. The patient treated with RT had 2 prior TSSs and received medical therapy.
Figure 1.
Treatment journey of the 18 patients on medical therapy at their longest follow-up. Each row represents the longitudinal treatment course of each patient before and/or after entering the study. Multiple boxes indicating medical therapy within the same patient represent different medications administered over time. Segments outlined in bold represent the follow-up period analyzed in the current cohort, from this’ study baseline to the longest available follow-up.
Abbreviations: CT, clinical trial; Keto, ketoconazole; Levo, levoketoconazole; Mety, metyrapone; Mife, mifepristone; Osilo, osilodrostat; Pasi, pasireotide.
Sixteen patients presented with recurrent disease; an additional 9 patients (13.4%) developed recurrent or persistent disease after surgery. HC replacement was administered at 21 of the longest available follow-up visits [6 due to ongoing hypopituitarism or adrenal insufficiency (AI) and 15 for temporary postoperative AI], with another 9 cases receiving replacement at intermediate follow-up visits.
All 18 patients on medical therapy at their longest follow-up received adrenal steroidogenesis inhibitors: osilodrostat (8 patients, 44.4%), metyrapone (6 patients, 33.3%), and ketoconazole (4 patients, 22.2%).
Comorbid Conditions
As shown in Table 2, mean body mass index (BMI) at baseline was 33.2 ± 7.6 kg/m2. Twenty-eight (41.8%) patients presented with DM, 15 (22.4%) with prediabetes, and 24 (35.8%) without DM. Fifty-five of 67 patients (82.1%) had hypertension at baseline. At the longest follow-up, mean BMI decreased to 30.6 ± 8.5 kg/m² (P < .001), and mean HbA1c decreased to 5.7 ± 0.9% (P < .001). Thirteen patients (19.4%) continued to have DM, and 9 patients (13.4%) had prediabetes. Hypertension was present in 35 patients (53.7%), of whom 25 (71.4%) were receiving at least 1 antihypertensive medication.
LNSC levels remained abnormal in 16 patients (23.8%), although LNSC data were not available for 21 patients (31.3%). Of those, LNSC testing was not considered clinically indicated in some cases, such as patients on HC replacement for postoperative AI (n = 10) or patients with adrenal CS status postadrenalectomy (n = 3). The remaining 8 patients with missing LNSC data were on medical therapy (n = 4) or status post-TSS (n = 4).
Cushing QoL
Sixty-five patients (71 baseline to follow-up case pairs) completed the CushingQoL assessment. In the overall cohort, treatment resulted in significant improvements in mean QoL scores at all follow-up time points: mean change in G1 was 16.6 ± 18.6 (P < .001); G2, 19.1 ± 19.4 (P < .001); and G3, 16.6 ± 27.1 (P = .009) (Table 3, Fig. 2A). For longest available follow-up for each case, overall mean improvement was 18.2 ± 20.9 points (P < .001).
Figure 2.
Score trajectory for (A) Cushing QoL, (B) BDI-II, (C) STAI-State, and (D) STAI-Trait in the overall cohort based on duration of follow-up, including patients with 2 follow-up visits. Significant improvements in mean scores were observed in all assessments and all follow-up time points except in group 3 STAI-State, noted with a gray line. Group 1: 6 months posttreatment, group 2: 12 to 18 months posttreatment, group 3: ≥ 24 months posttreatment.
Abbreviations: BDI-II, Beck Depression Inventory-II; QoL, quality of life; STAI, State-Trait Anxiety Inventory.
Table 3.
Cushing QoL scores at baseline, follow-up visit, and mean score change in each time-based group for total cohort, patients who had TSS and patients on medical therapy
| Category |
Subgroup |
n |
Baseline mean |
Follow-up visit mean |
Mean change |
SD (change) |
P-value |
| Total cohort |
Longest follow-up |
71 |
42.4 |
60.6 |
18.2 |
20.9 |
<.001 |
|
Group 1 |
45 |
40.6 |
57.2 |
16.6 |
18.6 |
<.001 |
|
Group 2 |
30 |
43.5 |
62.6 |
19.1 |
19.4 |
<.001 |
|
Group 3 |
23 |
41.2 |
57.9 |
16.6 |
27.1 |
.009 |
| TSS |
Longest follow-up |
42 |
40.0 |
59.9 |
20.0 |
18.5 |
<.001 |
|
Group 1 |
29 |
40.2 |
57.0 |
16.8 |
19.1 |
<.001 |
|
Group 2 |
21 |
41.4 |
61.9 |
20.4 |
15.8 |
<.001 |
|
Group 3 |
9 |
29.0 |
48.7 |
19.7 |
24.9 |
.045 |
| Medical therapy |
Longest follow-up |
19 |
46.3 |
58.4 |
12.1 |
26.2 |
.059 |
|
Group 1 |
9 |
44.6 |
56.7 |
12.1 |
18.5 |
.086 |
|
Group 2 |
7 |
40.9 |
57.1 |
16.3 |
31.4 |
.219 |
|
Group 3 |
10 |
56.0 |
62.0 |
6.0 |
27.9 |
.513 |
Abbreviations: QoL, quality of life; TSS, transsphenoidal surgery.
In the subanalysis by treatment strategy, 42 patients who completed the Cushing QoL achieved surgical remission and 19 patients were controlled on medical therapy. In the surgical cohort, improvement in scores were noted across all time groups with a mean score increase of 20.0 ± 18.5 points from baseline to the longest available follow-up (P < .001) (Figs. 3A and 4A). Among these patients, 15 had 2 follow-up visits; between them the mean score further increased by 9.6 ± 14.8 points, indicating significant QoL improvement >6 months postsurgery (P = .025). In contrast, patients under medical control at follow-up showed a mean improvement of 12.1 ± 26.2 points from baseline to the longest follow-up, which did not reach statistical significance (n = 19; P = .059) (Table 3, Figs. 3A and 4A).
Figure 3.
Mean score change in total cohort, patients after transsphenoidal surgery, and patients on medical therapy based on duration of follow-up: (A) Cushing QoL, (B) BDI-II, (C) STAI-State, (D) STAI-Trait. Direct comparison between the 2 treatment modalities was performed only in the longest available follow-up visit for each patient. Caps represent SEM. Only P-values ≤ .05 are displayed.
Abbreviations: BDI-II, Beck Depression Inventory-II; QoL, quality of life; STAI, State-Trait Anxiety Inventory.
Figure 4.
Mean scores at baseline and longest follow-up in total cohort, patients after transsphenoidal surgery, and patients on medical therapy for (A) Cushing QoL, (B) BDI-II, (C) STAI-State, (D) STAI-Trait. Caps represent SEM. Asterisks (*) indicate significant change from baseline to follow-up (P ≤ .05) and brackets significant differences between the 2 treatment modalities at the longest follow-up visit (P ≤ .05).
Abbreviations: BDI-II, Beck Depression Inventory-II; QoL, quality of life; STAI, State-Trait Anxiety Inventory.
MID achievement and predictors of improvement
In the overall cohort, CushingQoL MID was achieved in 42 of the 65 patients (64.6%) (Fig. 5). When stratified by follow-up duration, MID achievement rates were 60.8% in G1 (n = 45), 70.0% in G2 (n = 30), and 60.9% (n = 23) in G3.
Figure 5.
MID achievement rates for all patient-reported outcomes at most recent follow-up.
Abbreviations: MID, minimal important difference.
Males (n = 11) improved more than female patients (n = 54) (27.8 ± 13.0 vs 15.5 ± 21.9; P = .020) and achieved the MID more frequently (90.9% vs 59.3%; P = .045). Even though they presented with lower baseline scores compared to females (33.2 ± 16.3 vs 44.3 ± 20.7), that difference was not significant (P = .117).
Score change differed by BMI category, using as cut-off the baseline mean of our cohort (≤33.2 vs >33.2 kg/m²): patients with lower BMI (n = 34) improved considerably more than those with higher BMI (n = 31) (median score change: 26 vs 11; P = .023). Likewise, MID achievement was more common in the low-BMI group (76.5% vs 51.6%; P = .036).
Patients presenting with recurrent disease at baseline (n = 16) reported better baseline QoL than those with primary disease (n = 49) (51.6 ± 19.5 vs 39.5 ± 20.9; P = .046), and their mean improvement following treatment was smaller (7.2 ± 21.0 vs 21.0 ± 19.8; P = .022). Only 43.8% of recurrent cases achieved the MID compared to 71.4% of primary cases (P = .044).
Patients reporting symptom duration ≥3 years prior to diagnosis (n = 29) were less likely to achieve the MID compared to those with shorter symptom duration (n = 35) (48.3% vs 66.7%; P = .008).
Patients with at least 1 abnormal LNSC (n = 15) value at follow-up were less likely to meet MID compared to those with normal LNSC values (n = 28) (33.3% vs 75.0%; P = .008). Similarly, patients requiring HC replacement (after their first TSS or unilateral adrenalectomy for adrenal CS) for >6 months (n = 22) were more likely to achieve MID than those requiring ≤6 months (n = 30) (81.8% vs 50.0%; P = .019).
MID achievement rates between the TSS and medical-therapy groups differed (71.4% vs 47.4%) but did not reach significance (P = .070).
Baseline 24 hours UFC was inversely correlated with baseline CushingQoL score (ρ = −0.3; P = .035), indicating a relationship between biochemical and symptomatic disease severity.
BDI-II
Fifty-six patients (60 case pairs) were included in this subgroup. In the overall cohort, improvements in BDI-II score were seen at all follow-up time points: mean change in G1 was 4.7 ± 9.2 (P = .004); in G2, 7.7 ± 7.3 (P < .001); and in G3, 7.6 ± 10.6 (P = .008). In the overall cohort, mean improvement from baseline to the longest follow-up was 6.8 ± 8.6 points (P < .001) (Table 4, Fig. 2B). Of note, a significant 7.3-point improvement was noted between follow-up G1 (6 months) and follow-up G2 (12 months) (n = 11, P = .025), indicating continued improvement in depressive symptoms over time after treatment.
Table 4.
BDI-II scores at baseline, follow-up visit, and mean score change in each time-based group for total cohort, patients who had TSS and patients on medical therapy
| Category |
Subgroup |
n |
Baseline mean |
Follow-up visit mean |
Mean change |
SD (change) |
P-value |
| Total cohort |
Longest follow-up |
60 |
15.7 |
8.9 |
6.8 |
8.6 |
<.001 |
|
Group 1 |
37 |
17.0 |
12.2 |
4.7 |
9.2 |
.004 |
|
Group 2 |
26 |
15.2 |
7.5 |
7.7 |
7.3 |
<.001 |
|
Group 3 |
18 |
15.9 |
8.3 |
7.6 |
10.6 |
.008 |
| TSS |
Longest follow-up |
32 |
17.1 |
8.2 |
8.8 |
8.1 |
<.001 |
|
Group 1 |
22 |
18.6 |
13.6 |
5.0 |
10.9 |
.043 |
|
Group 2 |
17 |
14.7 |
6.7 |
8.0 |
8.1 |
<.001 |
|
Group 3 |
6 |
20.5 |
8.3 |
12.2 |
4.7 |
.001 |
| Medical therapy |
Longest follow-up |
18 |
14.4 |
11.0 |
3.4 |
9.9 |
.159 |
|
Group 1 |
8 |
14.6 |
11.0 |
3.6 |
6.7 |
.171 |
|
Group 2 |
6 |
18.3 |
10.8 |
7.5 |
7.1 |
.049 |
|
Group 3 |
9 |
11.8 |
8.8 |
3.0 |
13.3 |
.517 |
Abbreviations: BDI-II, Beck Depression Inventory-II; TSS, transsphenoidal surgery.
Among the 32 patients who underwent TSS, improvements were noted across all follow-up time groups, with mean scores decreasing from 17.1 ± 10.9 to 8.2 ± 7.0 at the longest follow-up (P < .001). In contrast, the 18 patients treated medically did not experience a significant change (P = .159). Improvement following TSS was significantly greater than with medical therapy at longest follow-up for each case (8.8 ± 8.1 vs 3.4 ± 9.9; P = .043) (Figs. 3B and 4B).
MID achievement and improvement predictors
Thirty-eight patients (67.9%) achieved MID by their longest follow-up (Fig. 5). Twenty-nine (51.8%) patients had baseline scores ≥14 points, indicating mild or moderate depression, and 23 (79.3%) of these patients met the MID. By follow-up duration, overall MID achievement rates were 56.8% in G1 (n = 37), 76.9% in G2 (n = 26), and 72.2% in G3 (n = 18).
By treatment approach, MID was met by 75.0% of patients who had TSS (n = 32) and 38.9% of patients on medication (n = 18) (P = .012). All patients who underwent BLA (n = 4) or RT (n = 1) and 5 out of 6 patients treated for adrenal CS achieved MID.
Patients with recurrent and primary disease did not differ in terms of baseline score (P = .267). However, those with recurrent disease were less likely to achieve MID (42.9% vs 76.2%; n = 14 vs 75.6%; n = 42, P = .021).
Symptom duration prior to diagnosis was inversely correlated with BDI-II score change (ρ = −0.33, P = .016). Patients experiencing symptoms for ≥3 years (n = 24) exhibited lower MID achievement rates compared to those with shorter symptom duration (n = 31) (50.0% vs 83.9%; P = .007).
Patients with normal LNSC at follow-up had higher MID achievement rates (81.5%; n = 27 vs 45.5%; n = 11, P = .026).
STAI
STAI-S
Fifty-six patients (60 case pairs) completed the STAI-State questionnaire. All follow-up time groups exhibited improvements, although in G3 the score decrease did not reach significance. In the overall cohort, mean scores declined from 44.8 ± 14.0 to 35.3 ± 11.2 at the longest follow-up (P < .001) (Table 5).
Table 5.
STAI scores at baseline, follow-up visit, and mean score change in total cohort, patients who had TSS and patients on medical therapy
| Outcome |
Category |
Subgroup |
n |
Baseline mean |
Follow-up visit mean |
Mean change |
SD (change) |
P-value |
| STAI-State |
Total cohort |
Longest follow-up |
60 |
44.8 |
35.3 |
9.6 |
12.5 |
<.001 |
|
|
Group 1 |
40 |
45.9 |
36.6 |
9.3 |
12.3 |
<.001 |
|
|
Group 2 |
25 |
46.2 |
35.3 |
10.8 |
10.8 |
<.001 |
|
|
Group 3 |
17 |
42.4 |
36.1 |
6.3 |
13.8 |
.078 |
|
TSS |
Longest follow-up |
33 |
44.4 |
34.3 |
10.1 |
12.3 |
<.001 |
|
|
Group 1 |
24 |
44.4 |
35.8 |
8.6 |
11.9 |
.002 |
|
|
Group 2 |
16 |
43.7 |
33.9 |
9.8 |
11.9 |
.005 |
|
|
Group 3 |
7 |
46.0 |
37.9 |
8.1 |
12.1 |
.126 |
|
Medical therapy |
Longest follow-up |
17 |
47.2 |
37.4 |
9.8 |
14.7 |
.014 |
|
|
Group 1 |
9 |
50.9 |
37.2 |
13.7 |
13.7 |
.017 |
|
|
Group 2 |
5 |
56.4 |
39.8 |
16.6 |
8.4 |
.012 |
|
|
Group 3 |
8 |
36.3 |
34.6 |
2.0 |
14.9 |
.715 |
| STAI-Trait |
Total cohort |
Longest follow-up |
58 |
46.0 |
37.3 |
8.6 |
12.6 |
<.001 |
|
|
Group 1 |
36 |
47.9 |
40.3 |
7.6 |
12.0 |
<.001 |
|
|
Group 2 |
26 |
45.7 |
36.0 |
9.6 |
10.9 |
<.001 |
|
|
Group 3 |
16 |
46.7 |
36.9 |
9.8 |
13.2 |
.010 |
|
TSS |
Longest follow-up |
31 |
47.5 |
36.7 |
10.7 |
12.2 |
<.001 |
|
|
Group 1 |
22 |
47.9 |
40.6 |
7.3 |
11.5 |
.008 |
|
|
Group 2 |
16 |
46.3 |
35.9 |
10.4 |
11.4 |
.002 |
|
|
Group 3 |
6 |
54.0 |
37.8 |
16.2 |
7.5 |
.003 |
|
Medical therapy |
Longest follow-up |
18 |
45.1 |
38.8 |
6.2 |
13.4 |
.065 |
|
|
Group 1 |
8 |
49.5 |
39.8 |
9.8 |
14.0 |
.089 |
|
|
Group 2 |
6 |
47.5 |
36.2 |
11.3 |
10.9 |
.052 |
|
|
Group 3 |
8 |
39.3 |
37.5 |
1.8 |
12.7 |
.709 |
Abbreviations: STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.
By treatment modality, state anxiety improved in both the TSS group (10.1 ± 12.3; n = 33; P < .001) and patients on medical therapy (9.8 ± 14.7; n = 17; P = .014) (Figs. 3C and 4C).
MID achievement and improvement predictors
Overall, 30 of 56 (53.5%) patients achieved MID in STAI-State at their longest follow-up visit (Fig. 5). By follow-up duration, MID achievement rates were 52.5% in G1 (n = 40), 56.1% in G2 (n = 25), and 64.7% in G3 (n = 17).
A negative correlation was observed between STAI-S score change and baseline age (ρ = −0.3, P = .029). Patients >40 years old at baseline (n = 29), improved less than younger patients (n = 27) [median score change: 5 vs 13 (P = .017)] and were less likely to meet the MID, with results approaching statistical significance (41.4% vs 66.7%, P = .058).
STAI-T
Fifty-three patients (58 case pairs) were evaluated. In the overall cohort, mean score change from baseline to longest follow-up was 8.6 ± 12.6 points (P < .001). In time-based subgroups the following score reductions were noted: G1: 7.6 ± 12.0 (P < .001), G2: 9.6 ± 10.9 (P < .001), G3: 9.8 ± 13.2 (P = .010) (Fig. 2D). Among patients treated with TSS (n = 31), significant improvement was seen in every subgroup. Patients receiving medical therapy (n = 18) showed numerical but not statistically significant improvement (P = .065) (Table 5, Figs. 3D and 4D).
MID achievement and improvement predictors
STAI-Trait MID was achieved by 28 (52.8%) patients at the longest follow-up (Fig. 5). By follow-up duration, MID achievement rates were 44.4% in G1, 53.8% in G2, and 68.8% in G3.
Patients ≤40 years at baseline (n = 26) improved more than those aged >40 years (n = 27), with results approaching significance [median score change: 14 vs 4 (P = .060)].
Patients with ≥2 Follow-up Visits
Twenty-eight patients had multiple follow-up visits; we stratified by follow-up duration (<2 years vs ≥2 years) [Table S1 (30)].
Cushing QoL
Significant improvements were noted in all groups with pairwise comparisons revealing higher scores in both first and second follow-up, with the mean score changing by 14.9 (P = .002) and 21.5 (P < .001) points, respectively, in total cohort.
BDI-II
Although the overall trajectory demonstrated significant improvement, pairwise comparisons showed no significant changes between baseline and first follow-up. Improvement was noted between baseline and the second follow-up visit (P < .001) and between the 2 treated visits (P = .021) (Table 6).
Table 6.
BDI-II mean scores and pairwise comparisons in patients with 2 follow-up visits
| Comparison |
Mean score A |
Mean score B |
Mean difference |
P-value |
| Baseline vs follow-up 1 |
16.9 |
13.0 |
4.846 |
.200 |
| Baseline vs follow-up 2 |
16.9 |
7.1 |
9.731 |
<.001 |
| Follow-up 1 vs follow-up 2 |
13.0 |
7.1 |
4.885 |
.021 |
Abbreviations: BDI-II, Beck Depression Inventory-II.
STAI-S
Overall, the mean score decreased from 45.9 ± 13.0 at baseline to 38.3 ± 12.4 at the first follow-up and to 36.1 ± 10.9 at the second follow-up (P = .005). In cases with follow-up ≥2 years (n = 13), the score trajectory did not change significantly from baseline (P = .187). In contrast, patients with total follow-up <2 years (n = 11) exhibited significant improvement (P = .008).
STAI-T
Overall, the mean score decreased from 49.2 ± 9.0 at baseline to 39.8 ± 11.6 at first follow-up and further to 36.4 ± 10.5 at second follow-up (P < .001). Significant improvement noted from baseline to both follow-up visits in both subgroups (P < .001).
Regression Analyses for Predictors of Change
In all measurements, after controlling for age and sex, baseline score was an independent predictor of greater change (P < .001) (Table 7). Patients with more impaired QoL, or severe depression and anxiety at baseline, had more room for improvement.
Table 7.
Predictors of mean score change from baseline to most recent follow-up of each patient in univariable and multivariable linear regression analysis
| Outcome |
Parameter |
Univariable analysis |
Multivariable analysis |
|
|
Estimate |
SE |
P-value |
Estimate |
SE |
P-value |
| Cushing QoL score change |
Baseline score |
−0.50 |
0.11 |
<.001 |
−0.47 |
0.11 |
<.001 |
|
Baseline age |
−0.05 |
0.20 |
.797 |
−0.04 |
0.19 |
.825 |
|
Male sex |
12.11 |
6.83 |
.081 |
7.49 |
6.68 |
.267 |
|
Baseline age ≤40 (vs >40) |
−3.43 |
5.23 |
.515 |
−4.90 |
4.89 |
.321 |
|
Normal LNSC (vs abnormal) |
−19.98 |
6.4 |
.004 |
−19.39 |
5.26 |
.001 |
|
HC replacement >6 months (vs ≤6 months) |
10.06 |
5.90 |
.095 |
12.35 |
4.96 |
.016 |
|
Primary disease at baseline (vs recurrent) |
−13.19 |
5.86 |
.028 |
−6.63 |
5.60 |
.241 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
−8.72 |
5.1 |
.095 |
−6.53 |
4.71 |
.171 |
|
Symptom duration ❤ years (vs ≥3 years) |
−4.60 |
5.25 |
.384 |
−4.55 |
4.70 |
.337 |
|
Treatment (TSS vs medical therapy) |
−7.87 |
5.8 |
.185 |
−4.23 |
5.41 |
.473 |
| BDI-II score change |
Baseline score |
0.57 |
0.09 |
<.001 |
0.58 |
0.09 |
<.001 |
|
Baseline age |
−0.08 |
0.09 |
.402 |
0.02 |
0.08 |
.797 |
|
Male sex |
−0.59 |
3.07 |
.848 |
0.80 |
2.53 |
.752 |
|
Baseline age ≤40 (vs >40) |
−3.96 |
4.82 |
.429 |
−0.52 |
2.02 |
.800 |
|
Normal LNSC (vs abnormal) |
−3.01 |
3.06 |
.332 |
−3.27 |
1.87 |
.090 |
|
HC replacement >6 months (vs ≤6 months) |
0.06 |
2.577 |
.980 |
2.33 |
1.90 |
.226 |
|
Primary disease at baseline (vs recurrent) |
−4.76 |
2.63 |
.076 |
−2.66 |
2.17 |
.224 |
|
Baseline BMI ≤33.2 kg/m2 vs >33.2 kg/m2 |
−3.79 |
2.29 |
.104 |
−1.41 |
1.90 |
.462 |
|
Symptom duration ❤ years (vs ≥3 years) |
−5.61 |
2.23 |
.015 |
−3.49 |
1.78 |
.055 |
|
Treatment (TSS vs medical therapy) |
−5.46 |
2.60 |
.041 |
−3.94 |
2.02 |
.057 |
| STAI-State score change |
Baseline score |
0.57 |
0.09 |
<.001 |
0.56 |
0.09 |
<.001 |
|
Baseline age |
−0.22 |
0.13 |
.104 |
−0.11 |
0.12 |
.338 |
|
Male sex |
−5.70 |
4.37 |
.197 |
−4.39 |
3.69 |
.239 |
|
Baseline age ≤40 (vs >40) |
−5.94 |
3.30 |
.078 |
−3.75 |
2.73 |
.175 |
|
Normal LNSC (vs abnormal) |
−2.15 |
3.95 |
.589 |
−4.47 |
2.89 |
.131 |
|
HC replacement >6 months (vs ≤6 months) |
0.72 |
3.45 |
.836 |
4.42 |
2.81 |
.123 |
|
Primary disease at baseline (vs recurrent) |
2.41 |
3.91 |
.743 |
2.14 |
2.91 |
.465 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
−2.36 |
3.38 |
.488 |
−0.93 |
2.56 |
.716 |
|
Symptom duration ❤ years (vs ≥3 years) |
−5.67 |
3.33 |
.095 |
−3.26 |
2.46 |
.192 |
|
Treatment (TSS vs medical therapy) |
−1.50 |
3.91 |
.970 |
−2.77 |
2.97 |
.355 |
| STAI-Trait score change |
Baseline score |
0.58 |
0.11 |
<.001 |
0.56 |
0.12 |
<.001 |
|
Baseline age |
−0.20 |
0.13 |
.128 |
−0.07 |
0.11 |
.562 |
|
Male sex |
−3.09 |
4.57 |
.502 |
−0.83 |
4.13 |
.841 |
|
Baseline age ≤40 (vs >40) |
−5.45 |
3.36 |
.111 |
−2.55 |
3.03 |
.405 |
|
Normal LNSC (vs abnormal) |
−6.52 |
4.23 |
.133 |
−6.74 |
3.44 |
.059 |
|
HC replacement >6 months (vs ≤6 months) |
4.63 |
3.52 |
.195 |
7.11 |
2.87 |
.018 |
|
Primary disease at baseline (vs recurrent) |
−2.07 |
3.90 |
.597 |
−0.34 |
3.42 |
.921 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
−4.95 |
3.38 |
.150 |
−2.59 |
3.00 |
.393 |
|
Symptom duration ❤ years (vs ≥3 years) |
−5.78 |
3.37 |
.093 |
−4.35 |
2.80 |
.127 |
|
Treatment (TSS vs medical therapy) |
−4.49 |
3.74 |
.236 |
−3.39 |
3.11 |
.281 |
Each predictor in multivariable analysis was separately evaluated, adjusting for baseline age, sex, and baseline score. In models exploring baseline age <40 years as a categorical variable, continuous baseline age was not included in the multivariable model. Statistically significant results (P ≤ .05) are indicated in bold.
Abbreviations: BDI-II, Beck Depression Inventory-II; BMI, body mass index; HC, hydrocortisone; LNSC, late-night salivary cortisol; QoL, quality of life; STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.
Cushing QoL
Normal LNSC at follow-up and >6 months of postoperative HC replacement were predictors of QoL score improvement and MID achievement even after adjustment for baseline score, age, and sex. Lower baseline BMI and male sex, although significant in univariable analysis, were no longer significant in the multivariable linear model. However, a BMI < 33.2 kg/m² (P = .034) and symptom duration ❤ years prior to diagnosis (P = .005) remained statistically significant predictors of reaching the MID in the multivariable logistic model (Table 8, Fig. 6). To determine if treatment modality modified the effect of LNSC, we built a model including baseline QoL score, age, sex, follow-up LNSC, and treatment type (TSS vs medical therapy). In this multivariable model, normal LNSC remained a significant predictor of improvement (P = .023).
Figure 6.
MID achievement predictors after multivariable analysis for (A) Cushing QoL, (B) BDI-II, (C) STAI-State. Each predictor was analyzed in a separate logistic regression model after adjustment for baseline score, age, and sex. Predictors for trait anxiety are not shown, as a longer duration of postoperative HC replacement was a significant predictor only in the linear multivariable regression model.
Abbreviations: BDI-II, Beck Depression Inventory-II; HC, hydrocortisone; LNSC, late-night salivary cortisol; QoL, quality of life; STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.
Table 8.
Predictors of MID achievement from baseline to most recent follow-up of each patient in univariable and multivariable logistic regression models
| Outcome |
Parameter |
Univariable analysis |
Multivariable analysis |
|
|
Estimate |
SE |
P-value |
Estimate |
SE |
P-value |
| Cushing QoL MID achievement |
Baseline score |
0.94 |
0.02 |
<.001 |
0.94 |
0.02 |
<.001 |
|
Baseline age |
1.01 |
0.02 |
.548 |
1.02 |
0.03 |
.410 |
|
Male sex |
6.89 |
1.09 |
.076 |
3.82 |
1.16 |
.249 |
|
Baseline age ≤40 (vs >40) |
1.01 |
0.52 |
.987 |
1.27 |
0.62 |
.704 |
|
Normal LNSC (vs abnormal) |
6.00 |
0.70 |
.011 |
22.82 |
1.17 |
.007 |
|
HC replacement >6 months (vs ≤6 months) |
4.50 |
0.66 |
.023 |
14.49 |
0.99 |
.007 |
|
Primary disease at baseline (vs recurrent) |
3.21 |
0.60 |
.050 |
1.78 |
0.68 |
.400 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
3.05 |
0.54 |
.039 |
4.33 |
0.69 |
.034 |
|
Symptom duration ❤ years (vs ≥3 years) |
4.29 |
0.56 |
.010 |
9.07 |
0.78 |
.005 |
|
Treatment (TSS vs medical therapy) |
2.79 |
0.57 |
.074 |
2.36 |
0.68 |
.209 |
| BDI-II MID achievement |
Baseline score |
1.08 |
0.04 |
.064 |
1.08 |
0.04 |
.042 |
|
Baseline age |
1.02 |
0.02 |
.510 |
1.01 |
0.03 |
.613 |
|
Male sex |
5.28 |
1.10 |
.130 |
5.76 |
1.14 |
.126 |
|
Baseline age ≤40 (vs >40) |
1.11 |
0.57 |
.854 |
1.05 |
0.63 |
.937 |
|
Normal LNSC (vs abnormal) |
5.28 |
0.78 |
.033 |
14.86 |
1.25 |
.030 |
|
HC replacement >6 months (vs ≤6 months) |
2.00 |
0.65 |
.288 |
2.32 |
0.71 |
.236 |
|
Primary disease at baseline (vs recurrent) |
4.27 |
0.65 |
.026 |
2.67 |
0.71 |
.165 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
1.94 |
0.58 |
.255 |
1.55 |
0.66 |
.504 |
|
Symptom duration < 3 years (vs ≥3 years) |
5.20 |
0.64 |
.010 |
5.74 |
0.70 |
.012 |
|
Treatment (TSS vs medical therapy) |
4.71 |
0.63 |
.014 |
4.19 |
0.69 |
.039 |
| STAI-State MID achievement |
Baseline score |
1.17 |
0.04 |
<.001 |
1.19 |
0.05 |
<.001 |
|
Baseline age |
0.97 |
0.02 |
.241 |
0.96 |
0.03 |
.261 |
|
Male sex |
1.95 |
0.71 |
.347 |
3.17 |
1.00 |
.249 |
|
Baseline age ≤40 (vs >40) |
2.83 |
0.56 |
.061 |
5.87 |
0.89 |
.048 |
|
Normal LNSC (vs abnormal) |
2.02 |
0.73 |
.337 |
2.41 |
1.04 |
.396 |
|
HC replacement >6 months (vs ≤6 months) |
0.94 |
0.59 |
.943 |
2.66 |
0.97 |
.313 |
|
Primary disease at baseline (vs recurrent) |
1.21 |
0.62 |
.757 |
2.15 |
0.92 |
.408 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
2.05 |
0.54 |
.189 |
1.57 |
0.82 |
.584 |
|
Symptom duration < 3 years (vs ≥3 years) |
1.39 |
0.55 |
.52 |
0.98 |
0.77 |
.980 |
|
Treatment (TSS vs medical therapy) |
1.95 |
0.62 |
.279 |
1.44 |
0.78 |
.634 |
| STAI-Trait MID achievement |
Baseline score |
1.17 |
0.05 |
<.001 |
1.17 |
0.05 |
<.001 |
|
Baseline age |
0.98 |
0.02 |
.295 |
0.97 |
0.03 |
.342 |
|
Male sex |
2.33 |
0.75 |
.257 |
4.16 |
1.02 |
.161 |
|
Baseline age ≤40 (vs >40) |
2.12 |
0.56 |
.175 |
2.32 |
0.76 |
.265 |
|
Normal LNSC (vs abnormal) |
1.78 |
0.71 |
.416 |
1.48 |
0.96 |
.686 |
|
HC replacement >6 months (vs ≤6 months) |
1.58 |
0.60 |
.450 |
4.21 |
0.95 |
.130 |
|
Primary disease at baseline (vs recurrent) |
2.45 |
0.61 |
.138 |
2.06 |
0.90 |
.421 |
|
Baseline BMI ≤33.2 kg/m2 (vs >33.2 kg/m2) |
1.98 |
0.54 |
.202 |
1.11 |
0.79 |
.891 |
|
Symptom duration < 3 years (vs ≥3 years) |
1.09 |
0.53 |
.866 |
0.99 |
0.71 |
.984 |
|
Treatment (TSS vs medical therapy) |
1.39 |
0.60 |
.585 |
1.18 |
0.82 |
.839 |
Each predictor in multivariable analysis was separately evaluated, adjusting for baseline age, sex and baseline score. In models exploring baseline age <40 years as a categorical variable, continuous baseline age was not included in the multivariable model. Statistically significant results (P ≤ .05) are indicated in bold.
Abbreviations: BDI-II, Beck Depression Inventory-II; BMI, body mass index; HC, hydrocortisone; LNSC, late-night salivary cortisol; MID, minimal important difference; QoL, quality of life; STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.
BDI-II
Symptom duration ❤ years (P = .012), normal LNSC at follow-up (P = .030), and TSS (P = .039) instead of medical therapy (for CD) were statistically significant predictors of MID achievement in the multivariable logistic models even after adjusting for age, sex, and baseline score (Table 8, Fig. 6).
STAI-S
In the multivariable logistic model adjusted for sex and baseline score, age <40 predicted higher odds of MID achievement (P = .041) (Table 8, Fig. 6).
STAI-T
After adjustments for sex and baseline score, age group <40 was no longer a predictor of improvement. Although nonsignificant in univariable screening, duration of postoperative HC replacement >6 months emerged as a significant predictor of score change, though not MID achievement, after adjusting for age, sex, and baseline score (Tables 7 and 8).
Discussion
In a clinical practice cohort of patients with CS followed prospectively before and over time up to 11.5 years after surgical remission and/or biochemical control from medical treatment, we identified significant improvements in mean QoL, depression, and anxiety scores in the overall cohort, but only half of patients achieved clinically meaningful improvements in anxiety, as assessed by MID, and about two-thirds of the cohort achieved clinically meaningful improvements in QoL and depression at their most recent follow-up. When assessed by treatment strategy, surgery resulted in statistically significant improvements in all 3 measures, whereas medical therapy resulted in statistically significant improvements in state anxiety but not QoL or depression. These findings may be impacted by the smaller cohort size of the medically treated patients and more complex treatment journeys in the medically vs surgically treated patients. Overall, in this cohort of treated, biochemically controlled patients, several predictors of improvements were identified, including age, baseline BMI, duration of symptoms prior to treatment, duration of HC requirement after surgery, and LNSC normalization with treatment.
PRO studies in CS have shown that patients with CS are at risk for mood disorders and impaired QoL at diagnosis and that improvement posttreatment is often partial, delayed, or inconsistent, even after biochemical remission (3-12). The most recent prospective study confirmed persistent deficits in QoL and depressive symptoms up to 1 year postsurgery, with mean BDI-II scores remaining in the clinically significant range (9). As for anxiety, a prospective study reported high baseline anxiety in patients with CD, and, although it improved after surgery, a proportion continued to experience anxiety up to 1 year posttreatment (14). Neuroimaging supports a biological basis for these symptoms, with brain abnormalities (hippocampal atrophy, cortical thinning, white matter damage) seen after biochemical cure possibly explaining the long-term emotional and cognitive deficits in some patients (12, 15). As for previously reported predictors of improvement, male sex, lower BMI at follow-up (4, 31, 32), LNSC normalization (17), and shorter duration of cortisol exposure (32, 33) emerged as independent predictors of better QoL. Persistent hormone deficits or arginine vasopressin deficiency were related to worse depression (9) while increased age and male sex predicted less anxiety (31). While some studies suggest that hypopituitarism and HC replacement are associated with poorer outcomes (11, 34), others found no significant difference (35). Limitations of these studies include the cross-sectional design (4, 31-36), small cohort sizes (9), and lack of long-term follow-up >12 months (37), especially in the setting of clinical trials (17).
In our study, QoL, depression, and anxiety improved following treatment, but the patterns varied by domain and follow-up duration.
As for QoL, interestingly, patients with recurrent disease showed better baseline QoL scores than those with primary disease, possibly due to posttreatment surveillance, resulting in earlier diagnosis at recurrence vs initial presentation. Although patients on medical therapy showed a trend toward improvement with treatment, results did not reach significance, potentially due to sample size or the increased (better) baseline scores in patients with recurrent disease and thus those receiving medical treatment. Most patients on medical therapy had persistent or recurrent disease and have experienced longer, more complex treatment journeys (as depicted in Fig. 1) compared to those in surgical remission, which also may impact QoL and mood outcomes. Notably, in patients with 2 follow-up visits, QoL continued to significantly improve 6 months posttreatment in those treated surgically but not in the total cohort.
Multivariable analysis revealed several predictors of QoL improvement after treatment. LNSC normalization was independently associated with approximately 20 times higher odds of achieving the MID, indicating the clinical importance of recovery of cortisol circadian rhythm for treated CS patients and the need for further work to identify medical therapies and regimens that can facilitate this. Postoperative HC replacement for more than 6 months after surgery (indicating a longer hypothalamic-pituitary-adrenal axis recovery) was also associated with greater QoL improvement. This finding complements prior work showing an association between duration of postoperative HC replacement and long-term remission (38, 39). Lower baseline BMI and shorter symptom duration were predictive of MID achievement, though not of mean score change.
As for depression, patients with 2 follow-ups had a distinct pattern: no significant change between baseline and first follow-up but significant improvement between the 2 follow-up visits. This suggests that depression may take longer to improve, with more evident change >6 months after biochemical control, which contrasts prior work suggesting that anxiety takes longer than depression to improve (14). The delayed trajectory could reflect the structural brain changes seen in CS even in remission, which are partially reversible (12, 40). Our data showed that symptom duration > 3 years prior to diagnosis reduced MID achievement, consistent with the literature linking diagnostic delay to persistent depression (33). A normal follow-up LNSC was associated with approximately 15 times higher odds of achieving the MID after adjustment, again emphasizing the need to attempt LNSC normalization while on medical therapy (9, 17).
As for anxiety, to date, no prospective study has assessed anxiety longitudinally using STAI, the gold standard for measuring and differentiating between trait and state anxiety (29). Our results confirm that anxiety improves after treatment; however, state and trait show different patterns. State anxiety was the only domain overall to improve significantly in the medical therapy group, while trait anxiety showed only a trend. Although age <40 predicted greater anxiety improvements in both, this remained significant only for state anxiety after adjustment in the logistic model. Trait anxiety improvements were predicted by longer postoperative HC replacement in the linear multivariable model, again suggesting that a shorter recovery time of the HPA axis may be an early indicator for identifying patients who require a closer follow-up. A normal LNSC at follow-up approached significance in the multivariable linear model, suggesting the importance of circadian rhythm recovery in trait anxiety improvement as well.
Across all measures, we found no baseline or outcome differences between pituitary and adrenal CS or between those on or off HC replacement at their last follow-up. Of note, our cohort was predominantly CD patients, and the small number of adrenal CS patients may limit the ability to detect a difference in the 2 cohorts.
Overall, discrepancies between mean change and MID achievement, as reflected in the linear and logistic models, respectively, highlight the importance of reporting both metrics when available, as they may capture different but clinically useful predictors.
We also observed differences between score change and MID achievement across different time groups within the same questionnaire. In STAI-State, G2 (12-18 months since most recent treatment) had greater score reductions than G3 (24 months or more posttreatment)—though change in G3 was nearly significant. However, a higher proportion of patients in G3 achieved MID. Looking at our data, G3 had the highest SD of mean change, indicating greater heterogeneity in treatment response, likely due to broader range of follow-up duration or higher medical therapy rates among patients: 45.5% (n = 10) in G3 vs 22.6% (n = 6) in G2% and 20% (n = 8) in G1. This variability in state anxiety is reflected in the subgroup of patients with 2 follow-up visits: those followed for >2 years showed no significant improvement, while those with <2 years did. Differential responses to long-term medical therapy, higher rates of loss to follow-up among postsurgical patients, or the negative impact of time on state anxiety symptoms may explain this. For BDI-II we used a percentage-based MID, which likely contributed to greater alignment with mean changes, and accounted for individual variability and baseline severity, factors especially relevant when applying generic tools in disease-specific contexts.
Of note, in the cohort overall, the mean follow-up score was within the normal range for depression (<14 for BDI-II) and anxiety (<40 for STAI) (41). This is an encouraging finding that, on average, patients with treated CS may have rates of depression and anxiety that are not clinically significant. Nevertheless, as shown in Table 2, rates of antidepressant, anxiolytic, pain, and sleep medication use did not decrease with treatment but instead were stable or increased numerically, although they were not statistically significant. Similarly, case-control studies have reported higher depression and anxiety levels in patients with CS in remission when compared to healthy controls, even if the mean scores were within the normal range for both groups (15, 42). Whether this difference is clinically significant still remains inconclusive. Taken together, these results emphasize the importance of multidisciplinary pituitary centers that integrate formal psychological services, including psychiatric care and social work support, to monitor and promote long-term mental health in this population.
Inclusion of both surgically and medically treated patients may be considered a limitation to the study, since it introduces heterogeneity in the cohort. However, including patients undergoing a range of treatments allows for analysis of CS cohorts as seen in a real-world practice rather than a controlled clinical trial setting, thus providing clinically valuable information. Another limitation of the study is the use of clinically available, rather than centralized, hormone assays, again introducing variability in our data. As this cohort included patients treated at our center, their endocrine testing followed standard of care, which did not include sending samples to a centralized laboratory. The use of antidepressants in a minority of patients could potentially affect depression scores. However, this is an unavoidable reality in patients with CS, and their use was stable over time (14.9% at baseline vs 19.4% at follow-up, P = .49). Given our prospective study design, which captured each patient’s change relative to their own baseline, and adjustment for baseline scores in multivariable models, any confounding is likely limited.
Despite these limitations, our data contribute to the literature as the largest clinical practice cohort to date that prospectively characterizes QoL and mood disturbances in CS patients, before and over time after achieving biochemical control. By incorporating 3 longitudinal time points, we identified that the greatest improvements occur within the first 6 months for QoL and anxiety, while depression improves more gradually beyond that point. Another strength of our approach is the use of score change and MID as outcomes when exploring potential predictors of improvement and not remission score per se, enabling more precise tracking of each patient’s progress and supporting an individualized approach by accounting for baseline severity.
In summary, this prospective analysis of mood and Qol in a clinical practice cohort of patients with CS showed that effective treatment of hypercortisolism improves depression, anxiety, and QoL, but one-third to one-half of patients do not experience clinically meaningful improvements in these measures. We identified predictors of improvement that highlight the need for early detection of CS and treatment strategies that allow for recovery of cortisol circadian rhythm. Psychological recovery in CS is heterogeneous, domain-specific, and not always aligned with biochemical normalization. Our findings support a model of care that extends beyond endocrine remission, integrating psychosocial follow-up and individualized treatment.
Acknowledgments
We would like to thank the people with Cushing’s syndrome who contributed their valuable time to this research.
Funding
This research was funded by the National Institutes of Health/National Cancer Institute Support Grant P30 CA008748.
https://academic.oup.com/jcem/advance-article/doi/10.1210/clinem/dgaf598/8307075?login=false
Filed under: Clinical trials, Cushing's, pituitary, symptoms | Tagged: depression, diabetes, hypertension, immune suppression, medication, mood, obesity, osteoporosis, post-op, surgery | Leave a comment »
CushieBlog 
